Distributed computing framework

Non-transitory computer-readable media storing instructions that, when executed by one or more processors, are configured to cause a controller server to perform operations comprising:

• receiving, from a seed node, information about initial conditions for a computational process representing an environment having one or more layers, wherein each layer of the one or more layers includes agents sharing a common characteristic that is different from characteristics shared by agents in other layers and the information includes initial conditions of the one or more layers;
• analyzing the information about initial conditions of the environment;
• in response to the analysis, determining a distribution of tasks corresponding to the computational process among one or more compute servers that are communicably coupled to one another, wherein each compute server of the one or more compute servers executes tasks to generate a portion of the environment;
• sending, to each compute server of the one or more compute servers, instructions for execution of tasks corresponding to a respective portion of the environment determined for the respective compute server;
• measuring, using a system clock, an execution time period for the computational process, wherein the execution time period is measured in units of discrete time frames;
• determining start of each time frame;
• at the start of each time frame, sending, to each of the one or more compute servers, an update instruction to update execution state of respective portion of the environment;
• receiving, from each of the one or more compute servers, a first response indicating completion of respective update;
• upon receiving a first response from each of the one or more compute servers, sending, to each of the one or more compute servers, a synchronization instruction to synchronize state of respective portion of the environment with other compute servers;
• and
• receiving, from each of the one or more compute servers, a subsequent response indicating completion of respective synchronization,
• wherein the operations generate, at a known stage of the computational process and in each iteration of the computational process, an identical result that is independent of a sequential order of receiving the first responses or the subsequent responses from the one or more compute servers.

A method comprising:

• receiving, at a controller server from a seed node, information about initial conditions for a computational process representing an environment having one or more layers, wherein each layer of the one or more layers includes agents sharing a common characteristic that is different from characteristics shared by agents in other layers and the information includes initial conditions of the one or more layers;
• analyzing the information about initial conditions of the environment;
• in response to the analysis, determining a distribution of tasks corresponding to the computational process among one or more compute servers that are communicably coupled to one another, wherein each compute server of the one or more compute servers executes tasks to generate a portion of the environment;
• sending, to each compute server of the one or more compute servers, instructions for execution of tasks corresponding to a respective portion of the environment determined for the respective compute server;
• measuring, using a system clock, an execution time period for the computational process, wherein the execution time period is measured in units of discrete time frames;
• determining start of each time frame;
• at the start of each time frame, sending, to each of the one or more compute servers, an update instruction to update execution state of respective portion of the environment;
• receiving, from each of the one or more compute servers, a first response indicating completion of respective update;
• upon receiving a first response from each of the one or more compute servers, sending, to each of the one or more compute servers, a synchronization instruction to synchronize state of respective portion of the environment with other compute servers;
• and
• receiving, from each of the one or more compute servers, a subsequent response indicating completion of respective synchronization,
• wherein an identical result is generated at a known stage of the computational process and in each iteration of the computational process, the identical result being independent of a sequential order of receiving the first responses or the subsequent responses from the one or more compute servers.